确诊病例
无症状感染者
上海疫情病例报告地点
人口密度图
疫情热力图

Abstract

System dynamics simulation is adopted to investigate how to maximize hunters’ gain from hunting foxes in a food chain that only consists of one producer trophic level and two consumer trophic levels, which are foxes and rodents, without bringing irreversible damage to the ecosystem of the island. This requires a balance between the number of consumers and the limitation of the food supply of the island ecosystem. To meet this requirement, the quantity of foxes to be hunted requires proper consideration according to the number of foxes and rodents and other factors. These factors for simulating the tendency of quantity variation of foxes and rodents include age composition and physiological phases composition of species, seasonal distributed birth rate, aging and disease, food demand, etc. A system dynamics model was constructed and series simulations and analyses were carried out. Based on the simulation results and analysis, 3 guidelines for hunting were proposed and verified, including avoiding overhunting, hunting in autumn and avoiding hunting infants.

通常使用特征描述子匹配的方式来实现图像检索，但是直接按照距离阈值匹配容易造成错误匹配。Wengang Zhou(周文罡) et al.(2010)根据特征描述子的的相对位置关系在不同图像中一致的特点，提出了 空间编码(Spatial Coding) 匹配算法，可以大幅提升检索性能。该方法在其他匹配场合也具有相当大的应用潜力。本文基于Matlab复现了该算法，并简单测试了该算法。

本文基于Mallat (1989) 提出的 Mallat 快速小波变换算法（Mallat 算法），使用 C++ 编程语言通
过 CUDA 平台复现了该 Mallat 快速小波变换算法，实现了对图像的多尺度分解与重构，并对算法
进行了并行优化处理，最后验证了算法的有效性并评估了优化效果。

使用 OpenMP 加速 HOG（梯度直方图）特征计算。 HOG 特征是数字图像处理领域一种经典的特
征描述子，其描述了图像中一部份区域的梯度方向和强度统计信息，在过去常常被用来结合 SVM（支持
向量机）进行行人检测等工作。本文基于 C 语言实现了 HOG 算法，并通过 OpenMP 进行了并行化加速
处理，最后通过 Python 接口展示算法结果，并对加速性能做了分析。结果显示使用 OpenMp 实现了有效
加速，当输入图像大于 500×500 像素时，使用 4 核处理器的加速比约为 3.72，加速效率约为 0.93。